This is a text2video model for diffusers, fine-tuned with a modelscope to have an anime-style appearance.
It was trained at 384x384 resolution.
It still generates unstable content often. The usage is the same as with the original modelscope model.

example images are here.